Application Of Bayesian Network Model To Analyze Prognostic Risk Factors Of Gastric Adenocarcinoma

Objective:The COX regression prediction model was used to analyze the risk factors for prognosis survival of gastric adenocarcinoma patients,and on this basis,a Bayesian network model was constructed to clearly and intuitively reveal the prognosis-related factors and potential network interactions among the factors,and to calculate and quantify the conditional probabilities among the variables,further exploring the practical value of Bayesian network model in clinical disease prediction and prognosis survival assessment,aiming to provide scientific basis and guidance for the diagnosis and treatment of gastric adenocarcinoma in the later stage of clinical practice.Methods:453 patients who attended the first hospital of Shanxi Medical University and were clearly diagnosed with gastric adenocarcinoma by pathological histological examination between January 2018 and January 2022 were selected and collected,and further patient-related data were reviewed in the medical record database,and variables including demographic characteristics,current medical history,treatment Status and pathological tissue subtypes were considered for selection,and 339 cases in the survival group and 114 cases in the death group were known after completing the follow-up.An Excel database was created to summarize the data and analyzed with the help of statistical methods,and finally,the meaningful variables were incorporated into a Bayesian network system to complete the drawing of the network topology,while the conditional probability values of each variable were calculated to infer the variable with the greatest strength of relationship with the outcome node.Results:After 453 patients with gastric adenocarcinoma were analyzed by K-M method survival analysis,the results showed that the overall survival was 4-56 months,and the median survival was 45 months(95% CI: 39.34-50.66);the screening results of one-way rank sum(Log-rank)test showed that age,tumor size,depth of tumor infiltration,lymph node metastasis,vascular invasion,distant metastasis,chemotherapy,lauren staging,clinical stage may be prognosis-related influencing factors for gastric adenocarcinoma(P < 0.05);The above variables were included in the COX regression model to complete the multivariate analysis suggested that the depth of tumor infiltration,clinical stage,lymph node metastasis,vascular invasion and chemotherapy were independent factors influencing the prognosis of gastric adenocarcinoma(P < 0.05);the results of mapping the Bayesian network model showed that the variables directly associated with the prognostic status of gastric adenocarcinoma were tumor infiltration depth,vascular invasion,and chemotherapy;while clinical stage could both directly correlate with prognosis and indirectly influence prognosis through tumor infiltration depth;tumor infiltration depth had a direct moderating effect on lymph node metastasis and jointly determined the prognostic survival time of patients with clinical stage.Conclusion:1.The depth of tumor infiltration,clinical stage,lymph node metastasis,distant metastasis,vascular invasion,and chemotherapy may be independent factors influencing the prognosis of gastric adenocarcinoma;2.With the help of Bayesian networks,it is possible to construct a risk factor prediction model related to the prognosis of gastric cancer and to accurately quantify the conditional probability values of each factor,which is a practical guide for clinical survival prediction;3.The factors directly related to the prognostic status of gastric adenocarcinoma are depth of tumor infiltration,vascular invasion and chemotherapy;while the clinical stage is both directly correlated with prognosis and indirectly correlated with prognosis mediated by depth of tumor infiltration;The depth of tumor infiltration has a direct moderating effect on lymph node metastasis and determines the survival time of patients together with the clinical stage.